An improved slime mould algorithm using multiple strategies

Abstract

Aiming at the defects of standard slime mould algorithm (SMA), such as local optima stagnation, slow convergence and improper balance between exploitation and exploration, we propose an improved SMA that contains the adaptive t-distributed variation strategy, improved location update formula and chaotic opposition-based learning strategy, that is, the MISMA. Utilizing comparative experiments and ablation studies on classical benchmark and CEC2020 benchmark suite, we proved that MISMA outperforms other state-of-the-art rival algorithms on convergence speed, solution accuracy, and robustness, each component of MISMA achieves an improvement on the defects of SMA at each stage and exhibits synergistic effects.